Skip to content

chore(context): free baseline word budget without dropping rules - #315

Closed
MendixMau wants to merge 4 commits into
masterfrom
chore/baseline-word-trim
Closed

MendixMau wants to merge 4 commits into
masterfrom
chore/baseline-word-trim

Conversation

@MendixMau

@MendixMau MendixMau commented Oct 13, 2026 •

Copy link
Copy Markdown
Owner

What

This frees always-loaded (baseline) skill text without dropping a rule. The baseline stood at 79,997 of its 80,000-word cap. This PR frees 3,738 words, leaving 3,741 words of headroom.

File Before After Δ
skills/conversion-runbook.md 15,216 14,336 −880
skills/agent-roles.md 5,372 3,057 −2,315
skills/source-triage.md 3,791 3,388 −403
skills/interview-protocol.md 3,367 3,227 −140
Baseline total 79,997 76,259 −3,738

Long incident narratives became one-line failure examples. Each one points at its section in the new on-demand process/baseline-incident-history.md, which keeps the original text verbatim. The runbook's copy of the open-questions status table now defers to interview-protocol.md §4. The six setup-only template shapes from agent-roles.md moved verbatim to agents/README.md, about 2,600 words; the installers never copy that file.

Every STOP row, rule, denominator, gate cell and one-line failure example stays in place. Review follow-ups:

  • the dispatch obligation pointer now leads to step 8a;
  • the stale "below" is fixed;
  • the history id is fixed;
  • the CHANGELOG figures are corrected.

Review

Independent review: MERGE-AFTER-SMALL-FIX, then MERGE.

  • 360 of 385 removed lines were found verbatim in the moved-to files.
  • The other 25 were read by hand. No protected text was lost, and all § pointers resolve.

Checks run

  • LC_ALL=C bin/render-routing.sh --check: Baseline 76259 words (budget 80000), rc 0.
  • test-surface-column 13/13, test-bug03-gates 31/31, test-obligation-fidelity 6/6, test-question-kinds 38/38, test-interview-gate 11/11, test-guide-reopen 10/10, test-triage-report 66/66.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw

The always-on baseline tier counted by bin/render-routing.sh --check was
at 79,735 of its 80,000-word cap. This frees 3,772 words (now 75,964)
by moving narrative and setup-only text out of baseline files, keeping
every rule, STOP, denominator and gate cell in place, each rule still
paired with a one-line failure example plus a pointer to the full text.

- conversion-runbook.md, source-triage.md, interview-protocol.md: long
  incident narratives -> one line + pointer; originals verbatim in new
  on-demand process/baseline-incident-history.md (R1-R20, AR1, ST1-ST4,
  IP1-IP3). Runbook's duplicate open-questions status table now defers
  to interview-protocol.md section 4.
- agent-roles.md: the six template shapes (setup-only) moved to
  agents/README.md, which installers never copy (explicit MXTK_AGENTS).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
…low", ST1 id

Independent review of #315: obligations.tsv still pointed at agent-roles.md for
step 8a, which moved to agents/README.md; agent-roles.md said the example bodies
are "below"; the history header cited a non-existent section id.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
Measured with LC_ALL=C on the branch after master (#313) was merged in:
79,997 -> 76,259 (-3,738); the old line used pre-#313 figures.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
MendixMau pushed a commit that referenced this pull request Oct 13, 2026
… rules

Review: MERGE — 360/385 removed lines found verbatim in moved-to files, other 25 read by hand; no protected text lost; baseline 79,997 -> 76,259.
MendixMau pushed a commit that referenced this pull request Oct 13, 2026
A feature PR does not squeeze other text to make room (the rule in #316).
Room comes from the baseline trim in #315, merged next. This branch's own
row stays trimmed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
MendixMau pushed a commit that referenced this pull request Oct 13, 2026
…6-fixes

Room in the baseline budget for this branch's detection-gaps row.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
MendixMau pushed a commit that referenced this pull request Oct 13, 2026
MendixMau added a commit that referenced this pull request Oct 14, 2026
)

Every constituent PR was reviewed by a fresh agent that did not write it (CLAUDE.md "Merging a PR"); only MERGE verdicts joined. CI green on each constituent head and on this head (40814fc); full suites run on the branch.

Review lines (per merged PR):
- #315 context trim — MERGE; 360/385 removed lines found verbatim in their new homes, rest read by hand; baseline 79,997 -> 76,259.
- #316 independent-review + word-budget rules — MERGE-AFTER-FIX (3 rounds), fixes applied as specified.
- #318 page-fidelity displayed text — MERGE after fix round + confirmation.
- #317 interaction-model skill (#291 part 1) — MERGE after fix round + confirmation; no client names leaked.
- #321 PROOF-OF-LOOK per section; final-look not due before derived Stage 6 — MERGE after fix round + confirmation.
- #307 graph-sweep guard; per-BRD conformance ledgers — MERGE after two fix rounds + confirmation.
- #325 extraction-quality section; token-burn fixes — MERGE after fix round + confirmation.
- #310 e2e OQL by alias, negative-span liveness, APP-DOWN, #302 credential regression fix — MERGE after fix round + 3 confirmations.
- #323 inbox promotions — MERGE after fix round + confirmation.
- #308 lint gate BLIND on reverted stock rules — MERGE (FIRST-SEEN reverted) after two fix rounds + confirmation.
- #328 pipefail + grep -q flake — MERGE; 0/30 flakes vs 15/100 on master. Follow-up: run-all resolves bin/lib/ subjects (reviewed, MERGE).
- #319 test-stack-up command-line ownership proof — MERGE after fix round + 2 confirmations.
- #327 change-intake core + on-demand detail; CRLF-safe change log — MERGE after fix round + confirmation.
- #320 follow-ups to #290/#292; Dispatching split out — MERGE after fix round + confirmation.
- #326 status delivery rounds (existing-app mode) — MERGE after fix round + confirmation.
- #312 lint batch 4: SEC014, toolkit CONV009 — MERGE after review + 2 fix rounds + confirmation; field run on mxcli 0.26.
- #324 exec.sh change classes, shared stale-lock takeover, SEC014 note echo — MERGE (joint review); follow-up owed for a takeover race after a mid-takeover crash.
- #322 MxTest after-startup guard, mxcli-run pid check — MERGE (joint review).
- #314 doctor property-set exit code; sync announces its bound — MERGE (joint review).
- #330 review follow-ups (exit 4/5 docs, pipefail, NUL strip) — MERGE-AFTER-FIX, fixes applied; takeover re-check dropped.

Field runs are still owed for most instruments in this batch (stated in each PR).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VJgWP5vEoAsNsJYqCDMGNw
@MendixMau

Copy link
Copy Markdown
Owner Author

Landed on master as part of the integration squash #329 (commit 31e90ae), after independent review — its Review: line is in that commit message. Closing as superseded.


Generated by Claude Code

@MendixMau MendixMau closed this Oct 14, 2026
@MendixMau
MendixMau deleted the chore/baseline-word-trim branch October 15, 2026 10:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants